It would be really nice if the SQL experience felt more like it was a natural part of the workflow.
Right now when you start a blank project, the flow experiences pushes you to first add a dataset. If that dataset is in a databases, you are asked to select your table to create your dataset. If you select writing with a custom SQL query, you are given a pretty distinct warning not to use this and a small box to enter some sql into. This most likely means you aren't starting here with SQL.
Another alternative for this is to just select a random table as the dataset so that you can then add a SQL recipe to your flow and work from there.
If you go to the notebook and start with a SQL notebook, you get a pretty nice SQL editor, with features like explain plans etc.
These experiences feel super disjointed and very confusing.
What would be nice is if these these experience were combined and we were able to start with the SQL notebook, write the SQL and then move from that view to back to the flow or into a Python/ R notebook.
It would also be great if we could take the SQL query, version control that into Github as a saved query for others to use.